{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:7QUKOTIXKX4CDNCIDPTQF4PFJM","short_pith_number":"pith:7QUKOTIX","schema_version":"1.0","canonical_sha256":"fc28a74d1755f821b4481be702f1e54b2e4d4582cd129557d1b5734e1a774328","source":{"kind":"arxiv","id":"2505.22002","version":1},"attestation_state":"computed","paper":{"title":"D-Fusion: Direct Preference Optimization for Aligning Diffusion Models with Visually Consistent Samples","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Fengda Zhang, Kun Kuang, Zijing Hu","submitted_at":"2025-05-28T06:03:41Z","abstract_excerpt":"The practical applications of diffusion models have been limited by the misalignment between generated images and corresponding text prompts. Recent studies have introduced direct preference optimization (DPO) to enhance the alignment of these models. However, the effectiveness of DPO is constrained by the issue of visual inconsistency, where the significant visual disparity between well-aligned and poorly-aligned images prevents diffusion models from identifying which factors contribute positively to alignment during fine-tuning. To address this issue, this paper introduces D-Fusion, a method"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2505.22002","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-05-28T06:03:41Z","cross_cats_sorted":[],"title_canon_sha256":"710653fd00f9f042c5f4fb5744f5c44ce529e8d4ee26c1827eaacb344aa7ba34","abstract_canon_sha256":"88201c57eb0796dadcf437b2c25fd4a008c0157efb73fa28c3f88c51c618a7d0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:11:07.694629Z","signature_b64":"8sTRFYgfGOuKLMglHc91JyrrHK6JuSFfkh+y1Z1GOpafbTzBiTnm7mjzv/qGHRjH4D7P/mv7CeaNwcQGEof4Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fc28a74d1755f821b4481be702f1e54b2e4d4582cd129557d1b5734e1a774328","last_reissued_at":"2026-07-05T11:11:07.694110Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:11:07.694110Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"D-Fusion: Direct Preference Optimization for Aligning Diffusion Models with Visually Consistent Samples","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Fengda Zhang, Kun Kuang, Zijing Hu","submitted_at":"2025-05-28T06:03:41Z","abstract_excerpt":"The practical applications of diffusion models have been limited by the misalignment between generated images and corresponding text prompts. Recent studies have introduced direct preference optimization (DPO) to enhance the alignment of these models. However, the effectiveness of DPO is constrained by the issue of visual inconsistency, where the significant visual disparity between well-aligned and poorly-aligned images prevents diffusion models from identifying which factors contribute positively to alignment during fine-tuning. To address this issue, this paper introduces D-Fusion, a method"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.22002","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2505.22002/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2505.22002","created_at":"2026-07-05T11:11:07.694179+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.22002v1","created_at":"2026-07-05T11:11:07.694179+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.22002","created_at":"2026-07-05T11:11:07.694179+00:00"},{"alias_kind":"pith_short_12","alias_value":"7QUKOTIXKX4C","created_at":"2026-07-05T11:11:07.694179+00:00"},{"alias_kind":"pith_short_16","alias_value":"7QUKOTIXKX4CDNCI","created_at":"2026-07-05T11:11:07.694179+00:00"},{"alias_kind":"pith_short_8","alias_value":"7QUKOTIX","created_at":"2026-07-05T11:11:07.694179+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.25759","citing_title":"Towards Anatomically Plausible Human Image Generation via Synthetic Localized Preferences","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2604.26341","citing_title":"SpatialFusion: Endowing Unified Image Generation with Intrinsic 3D Geometric Awareness","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2604.19406","citing_title":"HP-Edit: A Human-Preference Post-Training Framework for Image Editing","ref_index":13,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7QUKOTIXKX4CDNCIDPTQF4PFJM","json":"https://pith.science/pith/7QUKOTIXKX4CDNCIDPTQF4PFJM.json","graph_json":"https://pith.science/api/pith-number/7QUKOTIXKX4CDNCIDPTQF4PFJM/graph.json","events_json":"https://pith.science/api/pith-number/7QUKOTIXKX4CDNCIDPTQF4PFJM/events.json","paper":"https://pith.science/paper/7QUKOTIX"},"agent_actions":{"view_html":"https://pith.science/pith/7QUKOTIXKX4CDNCIDPTQF4PFJM","download_json":"https://pith.science/pith/7QUKOTIXKX4CDNCIDPTQF4PFJM.json","view_paper":"https://pith.science/paper/7QUKOTIX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.22002&json=true","fetch_graph":"https://pith.science/api/pith-number/7QUKOTIXKX4CDNCIDPTQF4PFJM/graph.json","fetch_events":"https://pith.science/api/pith-number/7QUKOTIXKX4CDNCIDPTQF4PFJM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7QUKOTIXKX4CDNCIDPTQF4PFJM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7QUKOTIXKX4CDNCIDPTQF4PFJM/action/storage_attestation","attest_author":"https://pith.science/pith/7QUKOTIXKX4CDNCIDPTQF4PFJM/action/author_attestation","sign_citation":"https://pith.science/pith/7QUKOTIXKX4CDNCIDPTQF4PFJM/action/citation_signature","submit_replication":"https://pith.science/pith/7QUKOTIXKX4CDNCIDPTQF4PFJM/action/replication_record"}},"created_at":"2026-07-05T11:11:07.694179+00:00","updated_at":"2026-07-05T11:11:07.694179+00:00"}